Re: RDF Schema media type

Hi Aaron,

On Mon, Sep 22, 2003 at 12:02:53AM -0400, Aaron Swartz wrote:
> Mark Baker proposes RDF Schema have its own media type to indicate that 
> RDFS inferences are necessary for proper understanding.
> 
> 1. RDF triples are supposed to be separable -- the rest should make 
> sense if any aren't understood. (This includes inferred triples, 
> obviously.)

Right, but I'm not sure how that relates to my issue.  I'm talking about
triples which aren't visible to a recipient or intermediary, but are
assumed visible by the sender.

> 2. Even if this was a good idea, the media type way of doing it isn't 
> scaleable.

If you mean registration of media types, then right, that isn't very
scalable.  But the "media type way" (i.e. message metadata by which
applications are dispatched) is a key element of Web architecture
that will always be required, because the same bits can mean different
things.

What's broken about media types is that they're not URIs.  That can be
fixed.  I'm working on it, in fact.  In the meantime, though onerous
(as you know 8-), a process exists for minting new ones.

> There are lots of different types of inferences.

Are there?  I don't know.  If there are lots, then the media-type-as-URI
solution would seem to be fairly high priority.

Can we at least agree that this is an issue; that RDF Schema isn't
self-descriptive?  I'm not detecting that you agree.

> For RDF Forms, you should probably just put RDFS inferences as a 
> requirement in the spec.

Yes, that's my plan, for now.  But many forms - namely those that don't
use "implicit intent"[1] - won't need inference, and it would be more
efficient to be able to make that call at design time rather than spec
time.

 [1] http://www.markbaker.ca/2003/05/RDF-Forms/#implicit_intent

Mark.
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca

Received on Monday, 22 September 2003 00:31:04 UTC